Chemical Characteristics of Aerosols in Coastal and Urban Ambient Atmospheres

Chemical characteristics of aerosols (PM1 (on-line measurement) and PM2.5 (off-line measurement) were investigated in coastal and urban atmospheres. Organics were the most dominant species in PM1 at both sites, exhibiting little difference in the relative fractions of chemical constituents in PM1 (organics, sulfate, black carbon (BC), nitrate, ammonium and chloride) between two sites. However, a clear difference in the types of organics was found between the two sites. The fraction of oxidized (aged) organics was much higher at coastal site than at urban site. The nitrate fraction significantly increased at both sites in PM2.5 compared with that in PM1, suggesting that a significant amount of nitrate exists at particle sizes of 1 μm–2.5 μm. Additionally, the aerosols observed at coastal site were acidic. At both sites, photochemical activity played an important role in enhancing sulfate and oxidized organics in the afternoon, thereby overcoming the dilution effect. More distinct diurnal patterns were observed for nitrate, BC and organics at the urban site compared to the coastal site. Chemical characteristics also varied with different air masses. The highest PM concentration was associated with the northwest continental air mass (the air mass passed over heavy industrial areas before arriving at the site and moved slowly compared to other air masses). Three PM events (sulfate-dominant versus organic-dominant) were observed during the sampling periods and were considered long-range transport (LTP) events. The water-soluble organic carbon (WSOC) and oxidized organic contents significantly increased during LTP events, suggesting that organics were highly aged during transport.


INTRODUCTION
Fine (< 2.5 µm) and ultrafine (< 100 nm) particles in the ambient atmosphere are of interest due to their effects on the Earth's radiation balance via scattering or absorbing solar light (Solomon et al., 2007), visibility impairment (Chow et al., 2002;Chang et al., 2009), cloud formation by acting as cloud condensation nuclei (CCN) (Solomon et al., 2007) and atmospheric chemistry (Reid and Sayer, 2002).Additionally, these particles influence human health (Dockery and Pope, 1994;Peters et al., 1997).They are directly emitted from various emission sources, such as industries, biomasses, vehicles and deserts (primary aerosols), or formed by gas-toparticle conversion processes (secondary aerosols) (Andreae, 2007).Due to their multiple sources, formation pathways and aging processes after formation (condensation, coagulation, oxidation and cloud processing), they exist in complex mixing states with multiple chemical constituents.
In urban environments, anthropogenic aerosols emitted from various local sources play important roles in air quality.Additionally, long-range transported aerosols can affect urban environments.In coastal and remote marine environments, natural aerosols (i.e., marine aerosols) are important because fewer anthropogenic aerosol sources are present in these areas; however, long-range transported aerosols can still affect the chemical characteristics and mixing states of aerosols in coastal and remote marine environments.Coastal sites are more affected by continental and anthropogenic sources compared to remote marine sites, leading to more complicated chemical characteristics and mixing states associated with submicrometer particles (Abbasse et al., 2003;Topping et al., 2004;Takami et al., 2005).Zorn et al. (2008) found a distinct difference in size distributions between oceanic air masses and continental outflow at coastal sites.A smaller peak in size distributions was observed for an ocean air mass (250 nm) than for a continental air mass (470 nm).Diesch et al. (2012) compared chemical characteristics of ambient aerosols in the coastal region of Spain.They found that chemical characteristics were significantly dependent on air mass types (i.e., long-range transported aerosols).Generally, the aerosol mass concentration was low and sulfate (not fully neutralized) became dominant when the air mass came from the ocean.
In this study, chemical components of non-refractory submicrometer particles in coastal and urban atmospheres were measured using a Q-AMS.Additionally, the black carbon (BC) and total mass concentrations of particulate matter less than 2.5 µm (PM 2.5 ) were measured at both sites.Filter samples were collected daily to determine the watersoluble inorganic ions (sulfate, nitrate, chloride, ammonium, sodium, potassium, calcium and magnesium), organic carbon (OC), elemental carbon (EC), and water-soluble organic carbon (WSOC) in PM 2.5 .These measurements enable us to compare detailed chemical characteristics of aerosols in coastal and urban environments and compare AMS chemical data (non-refractory submicrometer particles; online measurements) with filter-based PM 2.5 data (off-line measurements).An air mass backward trajectory analysis was also conducted to examine the effects of air mass types on the chemical characteristics of aerosols in coastal and urban environments.

EXPERIMENTAL METHODS
The ambient aerosol measurements were conducted in fall The coastal Boseong site is located at the Boseong Global Standard Meteorological Observation site (34.76°N,127.21°E), which is surrounded within 2.5 km by the Yedang plain and is near the South Sea (~1.6 km away), as shown in Fig. 1.The sizes of anthropogenic local sources in coastal Boseong should be small (the population in the ~660 km 2 area is approximately 45,000 and there are no large industrial complexes within 40 km of the area).The map of the urban Gwangju site is also shown in Fig. 1.Detailed descriptions of the Gwangju site can be found in a previous paper (Park et al., 2008).Briefly, the urban Gwangju site is influenced by a residential area, a commercial area, biomass burning from an agricultural area, an industrial area and traffic from a highway.
Table 1 summarizes the measured parameters of the ambient aerosols at the Boseong and Gwangju sites.Real- time measurements of organics, sulfate, nitrate, ammonium, chloride and MSA were made using the AMS (Aerodyne Research Inc., USA).The operation of the AMS has been described in a previous paper (Park et al., 2012).In brief, ambient aerosols were introduced into the instrument through a copper tube.All of the samples were passed through a PM 2.5 cyclone inlet (URG Corporation, USA) with a flow rate of 16.7 L min -1 to remove particles larger than 2.5 µm.Samples were then dried by a series of diffusion dryers before entering the AMS system.The sample flow was split into the AMS (0.07 L min -1 ).Isokinetic sampling was achieved by controlling the sizes of the sampling lines and flow rates.During the ambient measurement, the AMS was calibrated for electron multiplier (EM) gain, ionization efficiency (IE), and particle sizing based on the standard protocols (Jayne et al., 2000;Jimenez et al., 2003).IE calibration was conducted using size-selected ammonium nitrate particles (300 nm) (Sigma-Aldrich, USA), and the particle size calibration was performed using 90-250 nm polystyrene latex (PSL) particles (Duke Scientific, USA).For the mass concentration calibration, a particle collection efficiency (CE) of 0.5 was assumed to account for particle bounce losses in the AMS vaporizer (Alfarra et al., 2004;Takami et al., 2005;Takegawa et al., 2005).The relative ionization efficiency (RIE) values used in this study were 1.4 for organics, 1.2 for sulfate, 1.1 for nitrate, 1.4 for chloride and 4.0 for ammonium.The MSA fragment was added based on the laboratory standard fragmentation patterns of MSA in solution, as suggested in previous studies (Phinney et al., 2006;Zorn et al., 2008;Langley et al., 2010).Positive Matrix Factorization (PMF) was used to analyze unit mass resolution (UMR) of the AMS based on the mass spectrum of organics with an m/z below 150 using PMF2 software (Paatero and Tapper, 1994;Paatero, 1997).Before the PMF analysis, extremely high data points (i.e., spikes) were excluded from the time series of the mass spectrum, and an initial error matrix was modified following the procedure suggested by Ulbrich et al. (2009).A real-time measurement of the BC concentration was conducted using an aethalometer (Magee Scientific, USA), and the number size distribution of particles from 20 nm to 20 µm was measured using a scanning mobility particle sizer (SMPS) (TSI, USA) and an optical particle counter (OPC) (Grimm, Germany).
For determination of inorganic ions and WSOC in PM 2.5 , PM samples were collected on Zefluor filters (PALL Life Science, USA) through a URG cyclone (Teflon-coated aluminum cyclone with a cut size of 2.5 µm at 16.7 L min -1 ).To determine OC and EC, PM samples were collected on quartz filters (PALL Life Science, USA) using a high volume sampler (Thermo Electron Corporation, USA).The Zefluor filters were conditioned using a desiccator (temperature: 18 ± 2°C and relative humidity (RH): 20 ± 10%) for more than 24 hours before and after sampling.The quartz filters and aluminum foils were baked at 450°C for 4 hours to remove any remaining contaminants.
For inorganic ions analysis, PM samples were extracted with 20 mL of deionized and distilled water (DI, 18.2 MΩ) in an ultrasonic bath.The extracted solutions were filtered with a syringe filter (PTFE 0.45 µm) to remove any remaining particles before analysis using ion chromatography (Metrohm, Switzerland) along with a Metrosep C4-150/4.0column (1.0 mL min -1 flow rate, 100 µL injection volume and 4.0 mM HNO 3 eluent) for cations and a Metrosep A supp 5 column (0.7 mL min -1 flow rate, 100 µL injection volume, 1.0 mM NaHCO 3 /3.2mM Na 2 CO 3 eluent and 50 mM H 2 SO 4 suppressor solution) for anions.Five-point calibration curves were obtained using a standard solution before and after sample analysis.In the case of the WSOC analysis, the extraction procedure was the same as the inorganic ions analysis.Extracted solution was analyzed using a total organic carbon (TOC) analyzer (GE, USA).For the OC and EC analyses, a small portion (1.5 cm 2 ) of the quartz filter was removed and placed in the OC/EC analyzer (Sunset Laboratory Inc., USA) following the National Institute for Occupational Safety and Health (NIOSH) 5040 temperature protocol (Bae et al., 2004;Kondo et al., 2006).For quality assurance (QA) and quality control (QC) of data, 10% of filter samples were duplicated and variations among samples were found to be less than 5%.In addition, a good relationship was found between measured PM 2.5 mass and reconstructed mass obtained by summing mass concentrations of chemical components (r = 0.97).Meteorological data (temperature, wind speed, wind direction, and RH) at Boseong and Gwangju were obtained from the Boseong Global Standard Meteorological Observation site, which is located 10 m from the PM sampling site, and the Korea Meteorological Administration (KMA) site, which is located 50 m from the PM sampling site, respectively.
To determine the origins and pathways of air masses during the sampling period, an air mass backward trajectory analysis (NOAA HYSPLIT model; http://www.arl.noaa.gov/ready/hysplit4.html)was conducted (72 hours air mass backward trajectory analysis was performed at 500 m arrival heights above ground level at the sampling site).Aerosol Optical Depth (AOD) data from AERONET sun photometers (Level 2.0) in Korea were used to investigate the transport of aerosols over Korea peninsula (the AOD data were obtained from 6 sites in 2012 and 3 sites in 2013).2. Temperature and local wind speed were lower at the urban site than the coastal site during the given sampling period.At the coastal Boseong site, north and northwest winds were dominant, while northwest and northeast winds were prevalent at the urban Gwangju site.Time series of daily average PM 2.5 at Boseong and Gwangju are shown in Fig. 2. PM events were defined as periods when the daily PM 2.5 mass concentration exceeded the average value during the entire sampling period by more than one standard deviation (e.g., higher than 45 µg m -3 at Boseong).As shown in Fig. 2, two PM events (PM events 1 and 2) were observed at Boseong and one PM event (PM event 3) was observed at Gwangju.PM event 2 and PM event 3 occurred on the same day (11/23/2013), suggesting that regional aerosols led to enhanced PM concentrations at both sites.Air mass backward trajectory data and PM 2.5 variations over the regional scale can be used to infer characteristics of the PM events.All PM events observed were considered LTP events (i.e., long-range transported aerosols affected both sites).Simultaneous and regional increases in AOD and PM mass concentration over Korea peninsula during the PM events showed a clear evidence for the LTP events as shown in Fig. S1 in supplementary materials.Excluding data during the PM events, the average PM 2.5 concentration is 17.7 µg m -3 at Boseong, which is lower than the average value (22.7 µg m -3 ) at urban Gwangju.

Meteorology
PM 1 in this study is calculated by summing the concentrations of organics, sulfate, nitrate, ammonium, chloride, MSA measured using the AMS and the concentration of black carbon (BC) measured using the aethalometer.MSA was observed only in coastal Boseong (3.0%).Fig. 3 shows temporal variations of PM 1 (i.e., sum of organics, sulfate, nitrate, ammonium, chloride and BC concentrations) at (a) Boseong and (b) Gwangju.The PM 1 also increased during the PM events defined above.By excluding data during PM events, the average chemical characteristics can be compared between the two sites.At Boseong, the average PM 1 is 2.35 µg m -3 , and organic material was the most dominant, with an average concentration of 0.84 ± 0.57 µg m -3 , followed by sulfate (0.72 ± 0.39 µg m -3 ), BC (0.31 ± 0.20 µg m -3 ), nitrate (0.23 ± 0.18 µg m -3 ), ammonium (0.21 ± 0.23 µg m -3 ) and chloride (0.05 ± 0.05 µg m -3 ).At urban Gwangju, the PM 1 (10.21 µg m -3 ) was much higher than at Boseong.At Gwangju, organic material was also the most dominant, with an average concentration of 3.58 ± 3.48 µg m -3 , followed by sulfate (2.01 ± 1.39 µg m -3 ), BC (1.55 ± 1.30 µg m -3 ), nitrate (1.42 ± 1.34 µg m -3 ), ammonium (1.34 ± 0.95 µg m -3 ) and chloride (0.30 ± 0.37 µg m -3 ) at Gwangju.Both sites exhibited no significant difference with respect to the relative fractions of chemical constituents (organics, sulfate, BC, nitrate, ammonium, and chloride) in PM 1 .However, a clear difference in the types of organics was found between the two sites.At Boseong, HOA, SV-OOA and LV-OOA accounted for 21%, 26% and 53% of the total organic mass, respectively.OOA was more dominant than HOA, and LV-OOA was more abundant than SV-OOA.Additional  distinct constituents such as MSA-OA and PBOA, which can originate from biological marine organic materials, were not found.The percentage of LV-OOA was higher (53% versus 46%) and the percentage of HOA was lower (21% versus 39%) at coastal Boseong compared to urban Gwangju.This happened because the urban site had a plenty of combustion sources (e.g., traffic sources) leading to the higher HOA compared to the coastal site.Our data suggest that different types of organics (i.e., HOA and LV-OOA) are associated with distinct sources of PM 1 that differ between coastal and urban sites.Fig. 4 compares organics, sulfate, nitrate and ammonium concentrations of non-refractory submicrometer particles measured using an AMS at various remote marine and coastal sites around the world (Cavalli et al., 2004;O'Dowd et al., 2004;Topping et al., 2004;Quinn et al., 2004;Takami et al., 2005;Phinney et al., 2006;Zorn et al., 2008;Bates et al., 2005;Shank et al., 2011;Diesch et al., 2012).The non-refractory submicrometer concentration at the current coastal site was approximately 2 times smaller than the average of the other costal sites (7.30 µg m -3 ) and approximately 5 times higher than the average of the clean marine boundary layer sites (0.77 µg m -3 ).Sulfate was the most dominant species at remote marine sites in the Pacific Ocean, and organics increased at coastal sites.The mass fraction of the organics increased by up to 66% in the marine boundary layer during phytoplankton bloom periods, suggesting that high phytoplankton biomass in seawater may influence organics (i.e., biological species in seawater can be important sources of marine organic aerosols).At the coastal Boseong site, both organics and sulfate were equally dominant.
Diurnal variations of organics, sulfate, nitrate, ammonium and chloride at both sites were compared, as shown in Fig. 5.At both sites, organics, nitrate and BC were high in the morning, decreased in the midmorning, and remained at a lower level until evening (~16:00) when they increased again.The apparent midmorning decrease may be due to changes in the mixing layer height (increased dilution with increased mixing layer height).By contrast, sulfate increased in the afternoon.The sulfate increase is consistent with the LV-OOA increase (not shown here).Also, the time for increase of solar radiation and ozone were consistent with that for the enhanced sulfate.Strong photochemical activity could play an important role in enhancing the sulfate and LV-OOA concentrations in the afternoon by overcoming the dilution effect.More distinct diurnal patterns were observed for nitrate, BC and organics at urban Gwangju compared to patterns observed at coastal Boseong, suggesting that local urban sources such as morning and evening traffic also played a role in diurnal variations of organics, nitrate and BC.
During the sampling periods at both sites, two types of air masses were classified based on a cluster analysis of all air mass trajectory data: a north continental air mass (cluster I) and a northwest continental air mass (cluster II), as shown in Fig. 6(a).The air mass trajectories at both sites during the sampling periods were not significantly different.The north continental air mass (cluster I) originated from northern China or eastern Mongolia (45-55°N, 110-120°E) and arrived at the sampling site without passing over significant industrial areas in China or Korea.The northwest continental air mass (cluster II) passed over heavy industrial areas in China before arriving at the site, and the air mass moved slower than other air masses (i.e., it was more stagnant).The highest PM 1 was associated with the northwest continental air mass.Organics, sulfate, nitrate and ammonium increased 3-7 times compared to average values during the entire sampling period.However, the increase in BC was not as high as the other chemical components (i.e., the BC was less affected by the regional air mass trajectory compared to other chemical constituents).The mass concentrations of BC were 0.73 µg m -3 for the north continental air mass and 0.61 µg m -3 for the northwest continental air mass.The mass fractions of chemical constituents in PM 1 with different Chemical characteristics of PM 2.5 (filter-based) were also determined at the coastal Boseong and urban Gwangju sites.Excluding data during PM events, the average concentrations of chemical species in PM 2.5 at both sites are shown in Fig. 7.The PM 2.5 concentration was higher at urban Gwangju (20.5 µg m -3 ) than at coastal Boseong (15.7 µg m -3 ).At Boseong, sulfate (24.9%) and OC (23.2%) exhibited the highest contents followed by nitrate (19.3%) and chloride (9.5%).At Gwangju, OC (29.2%) was the most dominant species followed by sulfate (21.7%) and nitrate (21.3%).A higher sulfate fraction in PM 2.5 was observed at Boseong.In PM 2.5 , the nitrate fraction significantly increased at both sites compared with that in PM 1 , which was discussed in the previous section, suggesting that a significant amount of nitrate exists at particle sizes of 1 µm-2.5 µm.A comparison of anions (sum of sulfate, nitrate and chloride) and cations (sum of sodium, ammonium, potassium, calcium and magnesium) in PM 2.5 between sites (Fig. S2 in supplementary materials) suggests that aerosols observed at Boseong were more acidic than those observed at Gwangju.Chemical characteristics of PM 2.5 and PM 1 were determined during PM events, as shown in Table 3.Additionally, average values were included, excluding PM events.Three PM events (sulfate-dominant versus organicdominant events) were observed during sampling periods.All PM events were considered LTP events, and the northwest air mass was dominant (see Fig. 6(a)).In the LTP 1 event, sulfate was the most dominant species in both PM 2.5 and PM 1 .During the LTP 2 and LTP 3 events, where the longrange transported aerosols affected both sites on the same day, organics were the most dominant species.The OC/EC ratio also increased during the LTP 2 and 3 events.The WSOC in PM 2.5 and LV-OOA in PM 1 were much higher during the LTP events than during non-events, suggesting that organics were highly aged during LTP events.

CONCLUSIONS
The chemical characteristics of aerosols in coastal and urban environments were investigated using AMS chemical data (non-refractory submicrometer particles; online measurements) and filter-based PM 2.5 data (off-line measurements).Fractions of chemical constituent (organics, sulfate, black carbon (BC), nitrate, ammonium and chloride) in PM 1 were similar at both sites, although the PM mass concentration was much lower at the coastal site than at the urban site.However, the types of organics were clearly different between the two sites.The fraction of oxidized (aged) organics was much higher at the coastal site than that at the urban site.At both sites, the nitrate fraction significantly increased in PM 2.5 compared to that in PM 1 , suggesting that a significant amount of nitrate exists at particles of 1 µm-2.5 µm.Additionally, aerosols observed at the coastal site were found to be more acidic (nonneutralized) than those at the urban site.Diurnal patterns of nitrate, BC and organics were more distinct at the urban site than at the coastal site, and the enhanced sulfate and oxidized organics observed during the afternoon at both sites were likely caused by photochemical activity, overcoming the dilution effect.Aerosol characteristics also varied with different air masses.The PM concentration increased when the air mass passed over heavy industrial areas before arriving at the site (polluted air mass), and these air masses moved slowly compared to other air masses.Additionally, sulfate-dominant and organic-dominant PM events were observed at both sites.These events were LTP events in which organics were highly aged during transport.

Fig. 3 .
Fig. 3. Time-series of organics, sulfate, nitrate, ammonium, chloride, and BC concentrations at (a) coastal Boseong and (b) urban Gwangju, including pie charts of their average concentrations (data during PM events were excluded).

Fig. 6 .Fig. 7 .
Fig. 6.(a) Average air mass backward trajectories based on a cluster analysis (the shaded area represents major industrial areas near the air mass pathways), and (b) Pie charts of average concentrations of chemical constituents in PM 1 for the north continental air mass (cluster I) and the northwest air mass (cluster II).

Table 1 .
Measured parameters at the Boseong and Gwangju sites.

Table 2 .
Summary of the average temperature, RH, wind speed, and wind direction at the Boseong and Gwangju sites